Internet access is needed for initial downloads, model/runtime installation and update checks. Core local workflows can continue offline once the required components are present.
Frequently asked questions
Common questions about OpenMindAI.
Answers about privacy, offline use, models, storage, updates and hardware support.
Search every answer Platform-specific guidance
9 answers
The local-first design keeps supported local inference and conversation history on the machine. Network-connected features are separated from local execution and should clearly indicate when connectivity is required.
Models are stored in the storage root selected during setup rather than being committed to the application source repository. You can use a suitable local drive or fast external SSD with enough free space.
OpenMindAI's primary desktop path targets Windows 10 / 11, 64-bit. A modern x64 CPU and SSD storage are recommended, with 16 GB RAM recommended for the baseline local model experience.
The macOS installation path is designed for supported Intel and Apple Silicon systems. Package availability can vary by release, so use the exact assets and setup instructions published with the selected release.
OpenMindAI targets modern 64-bit Linux environments. Desktop/WebKit dependencies, package-manager differences and GPU/runtime stacks can vary by distribution, so follow the Linux installation guide for the release you use.
Use the APK that matches your device ABI when multiple packages are published, such as arm64-v8a, armeabi-v7a or x86_64. The Download Center lists every available package and its architecture dynamically.
Not always. Supported CPU execution is possible, while compatible GPU acceleration can improve local inference performance. Actual speed depends on model size, quantization, memory and backend support.
Yes. Published previous releases remain available in the release archive unless an administrator intentionally hides a release. Use the package and instructions that belong to the exact version you install.